Want to do some serious computing? Formerly a mainframe application found primarily in university and corporate computing centers, SPSS has succeeded in making the transition from a command-language application to an interactive, dialog-box- and menu-driven package for desktop computers. This powerful, option-rich program lets statisticians, market researchers, and scientists perform sophisticated statistical analyses on personal and corporate data sets. Version 10 improves on earlier versions, with new statistics options, customizable tool bars, interactive graphs and pivot tables, more-flexible output, and better handling of large data sets, although it comes at a higher price.
New Features Abound
The revamped Data Editor lets you enter and edit data and define variables in a single window. To generate statistics and charts, you choose menu commands and set dialog-box options. Although you’re still restricted to eight characters for variable names, you can now display descriptive labels.
All output is conveniently routed to a single, scrolling window. You can selectively show results and rearrange them via drag and drop. Version 10 lets you direct text output to the new Draft Viewer window, and you can export results to HTML. If you need to repeat an analysis, you can save the command syntax and run it from the new SPSS Production Facility. And you can create OLAP (online analytical processing) data cubes and slice them using any combination of values from your grouping variables.
Requirements and Limitations
The Mac version, which requires Mac OS 9 and at least a 233MHz G3, lags behind the Windows version in some respects. It won’t run as a client under the SPSS Windows NT server (a Mac server and an OS X version are being developed), OLE support is limited, and you can’t run multiple simultaneous sessions. And although it ran superbly on my G4, SPSS wouldn’t launch on an older Mac upgraded with a G3 card.
SPSS 10 is a dream program for statisticians, but it does require a statistics background. It won’t prevent novices from performing inappropriate analyses, confusing dependent and independent variables, or misinterpreting the results.
When generating an OLAP-cube report, you can choose any combination of grouping-variable values to display exactly the information you want.